A car enters the University of Chicago Medicine’s drive-through COVID-19 testing facility on March 20. 

The University of Chicago announced today that it is launching a COVID-19 medical imaging database as part of an initiative to help study the disease using artificial intelligence. 

The Medical Imaging and Data Resource Center (MIDRC) is funded by a two-year, $20-million federal contract from the National Institutes of Health. 

Over the next three months, researchers will upload more than 10,000 radiographs and CT-scans of COVID-19 patients to a database. The images will be shared with doctors and researchers from three professional organizations participating in the initiative: American College of Radiology (ACR), Radiological Society of North America (RSNA), and American Association of Physicists in Medicine (AAPM).  

“There currently are not enough curated data available to study. But having these top imaging organizations involved will make a difference — almost every scientist or clinician in medical imaging belongs to at least one of these organizations,” said Maryellen Giger, leader of the MIDRC and a radiology professor at the U. of C., in a statement. 

The ACR has recommended that medical imaging shouldn’t be used when diagnosing COVID-19 patients — the findings aren’t specific enough to distinguish it from other diseases. But the MIDRC will use artificial intelligence, training machine learning algorithms on its large dataset with the goal of improving diagnoses and treatments for patients. 

Over the past decade, artificial intelligence has become much more popular in radiology. It promises to move the field forward — as the authors of a 2018 review paper wrote, the introduction of AI could shift the discipline from “subjective perceptual skill to a more objective science.” 

That’s because, as with the MIDRC initiative, models are able to train themselves on thousands of data points, and are often capable of attaining insights about diseases that are difficult for human radiologists to achieve on their own. 

That comes with some dangers: a short report released this June found that neural networks trained on datasets with a gender imbalance will incorrectly classify the underrepresented gender more often when evaluating whether someone has a certain disease. While the result may seem intuitive, it highlights the importance of using datasets that are appropriately representative when calibrating artificial intelligence models. 

At a meeting earlier this year, researchers also highlighted the need for the Food and Drug Administration to adequately regulate the use of artificial intelligence technology in medical imaging. One participant pointed to the failure of computer-aided detection in breast cancer screenings — a 2015 study found that the technology could result in missed cancers. 

Still, the MIDRC could be useful in fighting the pandemic, which has persisted into the summer in Illinois and across the country. 

“Investigators will be able to access images and data to expedite research that will provide solutions to the COVID-19 pandemic,” Giger said. “This will speed up the sharing of new research on COVID-19, answering questions about COVID-19 presentation in the lungs, the efficacy of therapies, associations between COVID-19 and other co-morbidities, and monitoring for potential resurgence of the virus.”

Giger also said that the initiative could be expanded to other diseases at a later date.


Christian Belanger graduated from the University of Chicago in 2017. He has previously written for South Side Weekly, Chicago magazine and the Chicago Reader.

(0) comments

Welcome to the discussion.

Keep it Clean. Please avoid obscene, vulgar, lewd, racist or sexually-oriented language.
Don't Threaten. Threats of harming another person will not be tolerated.
Be Truthful. Don't knowingly lie about anyone or anything.
Be Nice. No racism, sexism or any sort of -ism that is degrading to another person.
Be Proactive. Use the 'Report' link on each comment to let us know of abusive posts.
Share with Us. We'd love to hear eyewitness accounts, the history behind an article.